Abstract
Anemia of CKD is clearly associated with reduced quality of life (QOL) and an increase in risk of death and cardiovascular events (2-4); this observation led to the hypothesis that aggressive correction of anemia might lead to improved mortality and better cardiovascular and QOL outcomes. Over recent years, average hemoglobin levels and EPO doses have been increasing, and higher targeted levels of Hb have been recommended in clinical guidelines. The rationale for these gradual increases in dose and targets was supported by extensive retrospective evidence (2-5) and QOL data that suggested a benefit to higher Hb levels.
Surprisingly, however, trial evidence has contradicted conclusions made from retrospective data. The Besarab trial, published in 1998 (6), suggested a potential cardiovascular risk from complete correction of anemia among ESRD patients, and three trials during intervening years have not demonstrated a significant benefit in terms of changes in left ventricular (LV) geometry for varying levels of anemia correction with EPO (7-9).
Anemia Management Trials
The Normal Hematocrit Study was a study of dialysis patients with cardiovascular disease published in 1998. It was controversial at the time as the study results contradicted conventional wisdom by suggesting that aggressive anemia correction may not be beneficial among hemodialysis patients with preexisting cardiovascular disease (6). The study examined the risks and benefits of normalizing the hematocrit in patients with cardiac disease who were undergoing hemodialysis. Over 1200 patients with clinical evidence of congestive heart failure or ischemic heart disease who were undergoing hemodialysis were included. Patients were randomized to EPO therapy with target hematocrit of 42% and 30% respectively. Median duration of treatment was 14 months and the primary end point was time to death or a first nonfatal myocardial infarction. Even though the prespecified stopping boundary had not been reached after 29 months, the study was halted at the third interim analysis on the recommendation of the independent data monitoring committee. The reported risk ratio for the normal hematocrit group compared with the low hematocrit group at the time of stopping was 1.3; 95% confidence interval (CI) 0.9 – 1.9. Much discussion occurred in the paper and in subsequent papers and presentations of the fact that, ultimately, the results of the study were not statistically significant. It is interesting to note though that the results only became nonsignificant after planned adjustment for multiple comparisons. The paper also included a post hoc analysis demonstrating an association between higher hematocrit and survival, and this may, it has been argued, merely have identified a subset of EPO-responsive patients with potentially better outcomes. The conclusions of the authors of the normalization of hematocrit trial were that administration of EPO to raise hematocrit to 42% was not recommended in patients with clinically evident congestive heart failure or ischemic heart disease who are receiving hemodialysis.
The Cardiovascular Risk Reduction by Early Anaemia Treatment with Epoetin Beta (Create) Trial
The CREATE trial randomly assigned 603 CKD patients from Europe, Asia, and Mexico to treatment with subcutaneous EPO-beta to achieve target Hb values of 13.0 – 15.0 g/dL or 10.5 – 11.5 g/dL. The primary end point was a composite of eight cardiovascular events. In this trial, complete correction of anemia did not affect the likelihood of a first cardiovascular event, as the difference between groups [58 events in the high Hb arm vs 47 in the low Hb arm; hazards ratio (HR) 0.78,95% CI 0.53 – 1.14; p = 0.20] was not significant. General health and physical function improved significantly in the high Hb group; however, dialysis was required in more patients in the higher Hb group compared to the low Hb group (127 vs 111, p = 0.03).
In summary, the results of the study suggested that targeting higher Hb was associated with improved QOL, a higher likelihood of requiring hemodialysis, and no appreciable increase in cardiovascular risk. When considering the prespecified primary outcome, the major criticism that has been leveled at CREATE is the fact that it was a somewhat underpowered study. The event rates were considerably less than those outlined in the prespecified power calculations. The prespecified anticipated event rate was 15%, yet the actual event rate of patients in the study was 6%, as acknowledged by the authors in the paper, and this lack of power is important when considering the numerical difference in events between groups in an overall context. It is also notable when comparing the CHOIR and CREATE studies that the event rate in the CHOIR study was approximately three times higher.
The Correction of Hemoglobin and Outcomes in Renal Insufficiency (Choir) Trial
The CHOIR trial (10) was an open-label trial of 1432 patients with CKD randomly assigned to receive a dose of EPO-alfa targeted to achieve a Hb level of 13.5 g/dL, or a dose targeted to achieve a level of 11.3 g/dL. Median study duration was 16 months and the primary end point was a composite of death, myocardial infarction, hospitalization for congestive heart failure, and stroke. The trial was stopped early by the independent data safety monitoring board because of a lack of conditional power to demonstrate an association between better outcome and therapy targeted to 13.5 g/dL and because of a potential risk in the high arm. There were 125 primary events in the high Hb group, compared with 97 events in the low Hb group (HR 1.34, 95% CI 1.03 – 1.74; p = 0.03). The major drivers of the result were the congestive heart failure hospitalization and mortality components of the end point, both of which strongly trended in the direction of increased risk among subjects in the high Hb arm. Quality of life improved over baseline, and assessments were similar in the two groups. The conclusion of the CHOIR trial, therefore, was an increase in cardiovascular risk without significant incremental QOL benefit with targeting of Hb to 13.5 g/dL compared with 11.3 g/dL.
A number of criticisms leveled at the CHOIR trial, including higher than expected dropout rates, higher dosing than in other trials, and small differences in baseline covariates, have been raised and discussed in other publications (11,12). The major reason for the high withdrawal rates was related to a prespecified requirement for withdrawal on initiation of renal replacement therapy. Withdrawal rates not including patients who required renal replacement therapy were much more comparable to similar clinical trials and to CREATE in particular. The dose of EPO used in CHOIR was also significantly higher than doses used in other recently published trials (7-9). Yet it is notable that the doses used were considerably less than the doses used in the USA-based Normalization of Hematocrit Trial among dialysis patients (6). The underlying reasons for higher prescribed EPO dosing among USA compared to non-USA patients in general remains unexplained, but it is important to note that both groups had higher doses than those required in non-USA studies with similar targets.
There were also minor differences in baseline covariates noted that are unlikely to have had a significant impact on results but have been raised as a potential concern (12), particularly in the light of a multivariate analysis performed and released by Ortho Biotech in their Clinical Study Report (13). Although point estimates for treatment effect were similar in direction and magnitude to those reported in the primary analysis in the CHOIR paper, the effect of treatment did not reach conventional statistical levels of significance following multivariable analysis. Interestingly, the two variables (among 100 or so other variables) that were different at baseline (i.e., history of coronary artery bypass grafting and self reported history of hypertension) were not included in the final multivariable model since they were not significantly associated with the composite events when other variables were in the model.
Three further relatively recent anemia management trials have focused on LV mass index as a surrogate outcome for cardiovascular disease (7-9). Roger et al. randomized 155 CKD patients with anemia to receive EPO-alfa as necessary to maintain Hb between 12 and 13 g/dL (group A) or between 9 and 10 g/dL (group B) (8). Patients were monitored for 2 years or until they required dialysis. The primary outcome was change in LV mass index at 2 years. Changes in LV mass index were not significantly different between the two groups, and the conclusion reached was that maintenance of Hb above 12 g/dL compared with 9 – 10 g/dL had similar effects on the LV mass index without affecting the development or progression of LV hypertrophy.
Canadian European Study
The Canadian European Study (7) was a double-blind study that randomized 596 patients that had recently initiated hemodialysis to Hb targets of 9.5 – 11.5 g/dL or 13.5 – 14.5 g/dL. Patients were treated with epoetin alfa over 24 weeks and were maintained for an additional 72 weeks. Patients with symptomatic cardiac disease or LV dilation were excluded. The primary outcome was LV volume index. Percentage changes in LV volume index between baseline and last value were similar (7.6% in the higher Hb and 8.3% in the lower Hb target group), as were the changes in LV mass index (16.8% vs 14.2%). In terms of QOL, SF-36 Vitality score was improved in the higher Hb versus the lower Hb target group. The conclusion reached was that, compared with partial correction, normalization of Hb in incident hemodialysis patients does not have a beneficial effect on cardiac structure.
The Canadian randomized trial of Hb maintenance to prevent or delay LV mass growth in patients with CKD was a 172 patient trial in which patients were randomized to maintain or achieve Hb level targets of 12.0 – 14.0 g/dL (120 – 140 g/L) or 9.0 ± 0.5 g/dL. There was no statistically significant difference between groups for the primary outcome of mean change in LV mass index from baseline to 24 months (9).
Peritoneal Dialysis (PD) Patients
A number of prospective randomized controlled trials of EPO therapy have included PD patients (14-17). These studies generally demonstrated that specific ESAs are effective in achieving target Hb levels among PD patients but did not examine mortality or cardiovascular outcomes. One study demonstrated improved QOL among all patients, including 51 PD patients, targeted to a Hb of 135 – 160 g/L compared to a Hb of 90 – 120 g/L (14). The small numbers of PD patients involved, however, preclude specific PD-related observations and it is not at all clear therefore whether PD patients should be treated differently than other CKD patients. As with other patients, retrospective data have demonstrated a correlation between anemia and both mortality and hospitalizations among PD patients (5), and Hb ≥ 12 g/dL has specifically been demonstrated to be associated with lower hospitalization rates (5) in adjusted multivariable analyses.
Peritoneal dialysis patients are less anemic and more sensitive to ESAs than their hemodialysis counterparts (18) and, in general, Hb levels achieved among USA PD patients are similar to those achieved among hemodialysis patients, but dosages among PD patients are remarkably less (19). While this is explained to some extent by the more frequent use of subcutaneous EPO among PD patients, as subcutaneous EPO is more efficacious (20), it may also be due to greater preserved renal function, less blood loss associated with the procedure, less iron deficiency, and less inflammation.
It is unlikely that randomized controlled trials specifically directed at Hb management among PD patients will be forthcoming and we are left therefore in the unfortunate situation of having to extrapolate from the data currently available for CKD and hemodialysis patients.
Commentary
In general, it can be concluded from these studies that ESA therapy to aggressive targets confers a potential mortality and cardiovascular risk. The QOL benefit appears to be reasonably consistent across studies and the fact that such a benefit was not observed in the CHOIR study should not alter the overall perception. Although CHOIR did not demonstrate incremental QOL improvement, QOL did improve from baseline in both groups over the course of the study.
There has been much discussion in the lay press and in the renal literature about the implications of these studies. The major consequence has been a fundamental shift in the management of anemia of CKD, such that the risks of aiming for higher Hb targets are being recognized. The April 2006 revision of the K/DOQI guidelines indirectly implied that an upper limit of 13 g/dL would be reasonable, and therefore seemed to support the observed trend toward increased ESA utilization in clinical practice. The response to the recently published data, however, has been to temper enthusiasm for Hb levels above 12 g/dL, and this is appropriate. Indeed, a new set of guidelines is being developed at time of writing and will probably be released prior to publication of this article. These changes in recommended targets are also important in the context of a recently issued (21) FDA “black box” warning with respect to ESA labeling in the future.
Aside from finally providing hard outcome data to guide the therapy of anemia, these studies are an important reminder to the nephrology community of the need for hard outcome trials to guide care. The body of evidence that supported Hb targets in the past was largely circumstantial in that it was retrospective in nature and was confounded by the observation that patients that are less anemic are generally not as unwell as those that are anemic.
It has become evident in recent analyses that patients that respond to EPO do better than those that do not respond, and one potential implication or corollary of this observation is the possibility that aggressive therapy in patients that do not respond may actually be harmful; this needs to be explored extensively in subsequent analyses.
Another important lesson from these trials, and particularly those with LV hypertrophy-related outcomes, is the fallacy of using surrogate outcomes to guide therapy.
The CREATE and CHOIR trials have not answered the fundamental question as to whether ESA therapy confers a survival or cardiovascular benefit at any level, and the ongoing Amgen-sponsored TREAT may shed further light on this issue despite the views of others that it might have been more appropriate to stop the study (22). Unfortunately, the TREAT upper target Hb of 13.0 g/dL means that a comparison of lower target Hb levels of 11 or 12 g/dL with the control arm of 9.0 g/dL will not be possible and thus leaves a burning and unanswered question.
Finally, with increasing emphasis on less aggressive targets and the dawn of multiple new ESAs on the market, it will also be important to generate good quality pharmacoeconomic data to guide clinicians in the optimal use of these expensive therapies. In considering the specific case of PD patients, while the greater efficacy of EPO could potentially lead to a tendency toward higher achieved Hb levels in this population, the absence of specific prospective outcome studies leads to reliance on extrapolation of data from other CKD populations, and therefore to a similar lack of enthusiasm for Hb levels above 12 g/dL.